Schizophrenia is a serious mental disease that costs $30 billion annually in the U.S. Treatment refractory patients account for the majority of these expenditures. Standard antipsychotic drugs are clinically effective in only 50% of the patients. The atypical antipsychotic clozapine (CLO) is the only treatment with proven efficacy in refractory patients. However, its use has been limited by its high cost and capacity to cause agranulocytosis. A second atypical drug risperidone (RIS) has not been adequately tested in refractory patients. Major questions facing the field are what should be the role of RIS in the treatment of refractory schizophrenia and what is its comparative efficacy to CLO. These questions are important in terms of clinical care and public health care allocation. Recent experience has demonstrated the poor understanding that clinicians and patients have concerning the appropriate uses of CLO and RIS, and the need for a study to determine their comparative efficacies in severely ill patients. The R-10 application to be conducted at three sites - Bronx Psychiatric Center (AECOM; J. Lieberman, P.I.); Rockland Psychiatric Center, and Manhattan Psychiatric Center in association with the Nathan S. Kline Institute for Psychiatric Research under the direction of Jan Volavka (P.I.) - describes a project that will address these questions. We will examine the comparative efficacy of CLO, RIS 6 mg/day, RIS 16 mg/day, and a standard agent, haloperidol (HAL) in severely ill treatment resistant patients with schizophrenia. Approximately 75 patients from each site (224 combined sample of all sites) will enter a 12-week clinical trial in which they are randomized to one of four treatments under double blind conditions. Doses of CLO and HAL will be titrated to target blood levels. Doses of RIS will be fixed at 6 or 16 mg/day. Patients' response in terms of symptom reduction will be assessed by PANSS, CGI, and Overt Aggression Scale, social functioning and activities of daily living by NOSIE, quality of life by the Quality of Life Interview, cognitive effects by a test battery, and side effects by ESRS. We hypothesize that the antipsychotic and the antiaggressive effects of CLO will be superior to RIS and HAL, and that RIS will be superior to HAL. RIS 16 mg/day will have more antiaggressive effect than 6 mg/day. The results of this study will provide the evidentiary basis for clinical practice and health care policy decisions.